Re: [PATCH] flush_cpu_workqueue: don't flush an empty ->worklist

[Oleg Nesterov <oleg@tv-sign.ru>]

On 01/09, Srivatsa Vaddagiri wrote:
>
> On Tue, Jan 09, 2007 at 07:38:15PM +0300, Oleg Nesterov wrote:
> > We can't do this. We should thaw cwq->thread (which was bound to the
> > dead CPU) to complete CPU_DEAD event. So we still need some changes.
> 
> I noticed that, but I presumed kthread_stop() will post a wakeup which
> will bring it out of frozen state. Looking at refrigerator(), I realize
> that is not possible. 
> 
> So CPU_DEAD should do a thaw_process on the kthread before doing a
> kthread_stop?

Probably we can do this, or that. In any case we need changes, that
was my point.

And the best change I believe is to _remove_ CPU_DEAD handling from
workqueue.c as I suggested before. This kthread_stop() is not a good
idea per se, it calls wake_up_process(), but we should in fact use
wake_up(&cwq->more_work). Yes, work->func() should be ready for the
false wakeups, but still.

But for now I hope the last "draft" patch is enough. I'll continue
on next weekend.

Oleg.